Re: P1788: Motion 45 PASSES
On 2013-08-03 17:15:12 +0200, John Pryce wrote:
> P1788
>
> According to our procedures, NO votes on actual standard text constitute a motion to amend. I believe two such motions arise from Motion 45:
> (a) on the status of Complete Arithmetic;
> (b) on the sign of an exact zero result, and related matters.
>
> The intent of the motion for (a) is quite clear. I propose this wording for it:
> ======
> Motion
> ======
> In last sentence of the text for Reduction Operations as approved by Motion 45, namely
>
> > It is recommended that these operations be based on an
> > implementation of Complete Arithmetic as specified in §13.7.
>
> the word "recommended" should be changed to "required".
> ======
>
> If anyone is unhappy with this wording, please say so. Otherwise, I
> will submit (on behalf of the Committee) it in a few days time.
This doesn't make sense, as complete arithmetic is only recommended.
Moreover, as already said, a standard mustn't require a particular
algorithm to implement some feature. A standard only describes the
behavior. You should change this to a note saying something like:
If Complete Arithmetic as specified in §13.7 is implemented
in hardware, it is recommended that these operations be based
on it.
FYI, even if Complete Arithmetic is implemented in software, an
implementation of the reduction operations may be faster with a
specific algorithm, not based on Complete Arithmetic. Indeed,
Complete Arithmetic first computes the exact result. But
reduction operations do not need this exact result in general.
Note also that you may be interested in reduction operations in
formats having a large exponent range, and again Complete Arithmetic
would be a bad solution.
--
Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <http://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)